Measuring Changes in Teenage Driver Crash Characteristics During the Early Months of Driving
This study involved in-depth examination of how rates of involvement in specific types of crashes change over time as new drivers gain experience, to improve our understanding of how driver education and graduated driver licensing systems could better prepare new drivers.
For media inquiries, contact:
Several studies throughout the world have documented that novice driver crashes decline sharply during the first 6 to18 months of driving, regardless of the age at which driving begins. It is clear that a substantial amount is learned during this period, but what that is has rarely been studied and remains largely unknown. The present study sought to shed light on how, and perhaps why, young novice driver behaviors change during the first few years by examining month-to-month changes in various crash characteristics, as compared to the overall pattern of declining crash rates.
North Carolina crash data from January 1, 2001 through December 31, 2008 were searched to identify crashes involving any person who had obtained an intermediate license (allowing unsupervised driving) at age 16 or 17 (N = 629,144). All crashes that occurred within the first 36 months after a teen obtained a license (N = 256,975) were included in the analyses. Plots of crash rates per licensed driver for each of the first 36 months of licensing were created, to examine whether certain crash types or characteristics declined more or less quickly. Data were also summarized in several other ways to provide detailed information about how young novice driver crash patterns change during the initial years of driving.
The large majority of crashes involved two vehicles and occurred while the young person was driving a car, on a roadway with a moderate posted speed limit (35 – 54 mph). Crashes occurring when the young driver was making a left turn or entering a roadway from a parking lot or driveway, as well as those in which the young driver’s vehicle overturned, ran off the road to the right, or hit a tree, utility pole, or legally parked vehicle declined at a particularly rapid rate. Additionally, crashes in which the young driver failed to yield, overcorrected or made an improper turn declined quickly. Crashes in which the young driver hit, or was hit by, another vehicle from the rear and those in which the novice was following too close declined more slowly than crashes overall.
The pattern of rapid decline in several crash types/characteristics strongly suggests these improvements were largely a matter of learning rather than some other process or a combination of processes. The few crash characteristics that declined more slowly over the first few years of driving appear to be partly the result of increased exposure, but they also reflect a tendency of young novices either not to allow sufficient headway or to be unable to react quickly enough when a leading vehicle slows or stops. If learning is involved in the decline of these kinds of incidents, it is offset to some extent by other factors. The appearance of rapid learning about some seemingly straightforward driving maneuvers or situations, among drivers who had been required to drive while supervised for 12 months before beginning to drive on their own, was surprising.
This study provides an extensive and detailed look at crashes during the first few years of driving during which dramatic improvement occurs. Such analyses are not possible in most jurisdictions, either because the necessary data do not exist or the samples are too small to identify clear month-to-month patterns. The exploratory nature of this study, along with some inherent limitations of crash report data, preclude firm conclusions. Nonetheless, there are some hints for how parents might better assist their teens in learning to drive. The findings also raise some questions that other research approaches – especially naturalistic driving studies – may be able to address more directly than analysis of crash report data.
For media inquiries, contact: